Recent advances in hardware and communication technologies have brought about an era of client computing over wired and wireless networks. The proliferation of powerful notebook computers and wireless client devices promises to provide client end users with network access at any time and in any location over various networks, including the Internet. This continuous connectivity allows users to be quickly notified of changing events, and provides them with the resources necessary to respond in realtime even when in transit. For example, in the financial services industry, online traders and financial professionals may be given the power to access information in real-time, using wireless client devices.
Conventionally, however, developers of complex, wireless messaging solutions have been forced to develop applications that are specific to various device types and network access protocols in diverse enterprise architectures and platforms. In other words, conventional client computing solutions have been largely platform-specific, network-specific, or both. For example, messages may be generated by a wide variety of applications running on a wide variety of client devices, such as Palm computing platform devices, Windows CE devices, paging and messaging devices, laptop computers, data-capable smart phones, etc. Depending on the type of network used by service providers, the client-generated messages may be transported over networks having various access protocols, such as, e.g., Cellular Digital Packet Data (CDPD), Mobitex, dial-up Internet connections, Code Division Multiple Access (CDMA), Global System for Mobile Communications (GSM), and ReFlex. As a result, current developers of client computing solutions must have intimate knowledge of specific network characteristics including, e.g., wireless network characteristics, protocol environments, and wireless links channel characteristics. Therefore, there exists a need to simplify wireless client and server application development environments to support the wide variety of device and network dependent architectures.
Messaging Application Programming Interface (MAPI) is a messaging architecture and an interface component for applications such as electronic mail, scheduling, calendaring and document management. As a messaging architecture, MAPI provides a consistent interface for multiple application programs to interact with multiple messaging systems across a variety of hardware platforms. MAPI provides cross platform support through such industry standards as Simple Mail Transfer Protocol (SMTP), X.400 and common messaging calls. MAPI is also the messaging component of Windows Open Services Architecture (WOSA).
Accordingly, MAPI is built into such operating systems as, e.g., Windows 95, Windows 98, Windows NT and Windows 2000, available from Microsoft Corporation of Redmond, Wash., U.S.A. and can be used by 16-bit and 32-bit Windows applications. For example, a word processor can send documents and a workgroup application can share and store different types of data using MAPI. MAPI separates the programming interfaces used by the client applications and the service providers. Every component works with a common, Microsoft Windows-based user interface. For example, a single messaging client application can be used to receive messages from fax, a bulletin board system, a host-based messaging system and a LAN-based system. Messages from all of these systems can be delivered to a single “universal inbox.”
Transmission Control Protocol (TCP) is a transport layer protocol used by an application in one host to communicate with an application in another host. This is accomplished by services provided by the protocol layers beneath the transport layer in both hosts. As a connection-oriented protocol TCP requires the establishment of a connection between the two hosts before two applications are able to communicate. TCP manages the connection and once both applications have communicated all required information between themselves the connection is released as if the connection is two simplex connections as opposed to a single duplex connection. The information is transferred between applications on different hosts is a byte stream. The transport layer hides message transfer details such as segmentation, ordering and duplication from the applications and provides end-to-end acknowledgement.
The Internet Protocol (IP) layer provides certain services to the transport layer protocol including hiding the details of the physical and data link layers and the services provided by them from the transport layer protocol. The IP layer provides a datagram delivery service. A datagram is a unit of data less than an entire message. A message may be, for example, a file, which may be quite large. Since there is a maximum size for a message (or file), the message may have to be segmented and transferred in smaller units. These smaller units are thus called datagrams. Each datagram is sent over the network as a separate entity and may, in fact, follow separate paths to the destination host. At the destination host, the datagrams are reassembled in proper order (usually in a buffer area) by the transport layer. Each node on the network sends any datagrams on to a next node only considering the final destination and only acknowledges receipt of the datagram to the preceding node. That is, the IP layer does not provide end-to-end acknowledgement. End-to-end acknowledgement is a service of the transport layer protocol. Should any node-to-node acknowledgements not be received by the preceding node, the datagram or datagrams unacknowledged will be retransmitted. The transport layer in the destination host will also acknowledge any duplicated datagrams (else receipt of duplicate datagrams will continue resulting in a clogged network) and ignore them.
Routing between network nodes is accomplished by means of muting tables. These routing tables can be static or dynamic and result in datagrams being forwarded from a source host to a destination host one node at a time. The intermediate nodes are often called “hops.”
The acronym, TCP/IP, is also used to refer to a five layer protocol model similar to the ISO/OSI seven layer protocol model. The TCP/IP model does not have the equivalent to layers 5 and 6 of the ISO/OSI protocol model. A protocol model defines the protocol layers and the interfaces between the layers. When implemented in software, hardware or firmware or possibly field programmable gate arrays (FPGAs), the implementation provides the actual services. This layered approach allows for ease of upgrading so long as the interface to the layer immediately above or below is not altered. Layering also allows for complete substitution. For example, should a new physical medium become available then so long as the interface between layer two and layer one remain the same, an old physical layer implementation module can be removed and a new implementation module substituted. In the alternative, the new implementation module could be added as another option. That is, the protocol suite defines the rules and the implementation provides the services that allow the communications to take place between applications on different hosts. The implementation of the TCP layer provides for the application to require a certain Quality of Service (QOS) as specified by a set of parameters including but not limited to priority, transmission delay, throughput etc.
Another well-known transport layer protocol is known as User Datagram Protocol (UDP), which is a connectionless transport protocol. The basic data transfer unit is the datagram. A datagram is divided into header and data areas as it is for the IP layer. An additional header over and above the IP header is used. The UDP header contains source and destination addresses (ports), a UDP length field (the length includes the 8 byte header and the data) and a UDP checksum. The UDP data includes the IP header and data. The IP layer supports UDP as a connectionless transport protocol for use in transmitting, for example, one time request-reply type messages or applications where time is of greater importance than accuracy.
TCP is used by applications on different hosts to communicate over an assumed unreliable network. To accomplish such communication much is added to the protocol in order to ensure that the information transferred over the network is valid. This addition has a cost and that cost is increased overhead with the attendant increase in bandwidth. A UDP header is eight bytes, the TCP header is 24 bytes and an IP header is a minimum of 20 bytes. Therefore, UDP/IP headers are a minimum of 28 bytes and TCP/IP headers are a minimum of 44 bytes. This is fairly large in terms of overhead and bandwidth utilization particularly over wireless networks. There are other significant problems with using standard TCP/IP over wireless networks principally in the area of flow control. The UDP/IP protocol combination, while not offering any guarantees to users, is expected to be reliable. Wireless networks tend, however, by their nature to be lossy. Several. solutions have been proposed when the network is not homogeneous. That is, when the network has both wireless and wireline portions. One suggestion is to use indirect TCP and another is to use snooping.
Other protocols such as Serial Line IP (SLIP) and Point to Point Protocol (PPP) have been developed. SLIP is not a standard and both are for point to point connections only so are not available for uses in networks. CDPD vendors indicate that they provide an integrated TCP/IP stack but it is not known the cost in terms of bandwidth overhead.
Conventionally, the existing wireless protocols do not provide an end-to-end solution over multiple networks and multiple client devices. Therefore, in addition to the need for a common architecture through a single, user friendly methodology for providing effective and reliable wireless data solutions for hand-held and laptop computing devices, wireless networks, and legacy systems, there also exists a need to efficiently and reliably communicate data using minimum bandwidth.